OBJECTIVE: Egocentric social network analyses show that drinking habits of college students' friends predict personal alcohol consumption. To date, most of this research focused on between-person, cross-sectional, or long-term longitudinal designs to evaluate these effects. This study used intensive longitudinal methods to examine episode-specific effects of social networks (network drinking, network composition) on college students' drinking, comparing within-person and between-person effects on individual episodic drinking, and highlighted social network characteristics that might be targeted for intervention. METHOD: College students ( = 1,151, 54% female, = 19.26, 81% White) identified their ego network: five close alters (e.g., friends, family) whom they met frequently. For 30 days using an internet-based diary, participants reported their daily drinking levels, indicated whether they met with each alter the previous night, and reported how many drinks each alter consumed. RESULTS: Two-part multilevel models were used to examine participant drinking. At the within-person level, when a higher proportion of a participant's ego network drank and when network total drinks were higher than average, the odds that a participant drank were significantly increased, and if they did drink, they consumed more alcoholic drinks than usual. Participants were more likely to drink on days when their network was comprised of more friends and more opposite-gender alters. CONCLUSIONS: Findings provide preliminary evidence that daily network drinking appears to influence drinking patterns and quantities among students. Such information could be used within ecological momentary interventions to prevent harmful drinking patterns such as heavy drinking episodes and incapacitation. (PsycInfo Database Record (c) 2024 APA, all rights reserved).